Adaptive Determination of Time Delay in Grey Prediction Model with Time Delay

Received: 18 March 2019 Accepted: 1 August 2019 This paper attempts to determine the time delay between the sequence of system behaviors and that of influencing factors on system behaviors. Firstly, a new concept was coined called the representative subsequence (RS), and the RSs of influencing factors were selected. Next, a new geometric similarity grey relational grade (GRG) model was set up to compute the time delay of each influencing factor RS relative to the sequence of system behaviors. On this basis, the time-delay corresponding to the maximum GRG was taken as the desired time-delay. The verification on actual data shows that the concept and extraction rule of the RS are feasible, and our geometric similarity GRG model outperforms the conventional model in predicting time delay. The research findings lay the basis for grey prediction modelling of time sequences and optimizing the prediction effect of grey prediction models.